New trends in passive localization by multiarray network

We focus on the problem of passive localization of radio-sources, solved by means of several separated multi-arrays (base stations) linked with a central process unit. Traditionally, the procedure relies on two distinct steps: intermediate parameters are first estimated (like angles-of-arrival or times-of-arrival), then the position is deduced thanks to these transmitted intermediate parameters. New strategies estimating the position directly by considering the whole base station network as a unique sensor network appeared recently and offered promising results. To the best of our knowledge, the characterization of their performances has not yet been established in a deterministic signal context. In this article the model is developed and the corresponding Cramer-Rao bound is expressed. This bound represents a useful tool to quantify what new one step approaches can offer in terms of optimal performances. Numerical study compare here the proposed bound to the performances of traditional 2 steps and recent one step approaches. We show that for one source existing techniques performs well and equivalently, but the two sources case underline the gap between 2 steps and 1 step approaches. Moreover, we underline the role played by the time-bandwidth product.